0:00 Good afternoon everyone. My name is Will Samosir and I’m the CTO and co-founder for VARIANT3D. We are a knitting technology company based in Malibu, California. Founded by Garrett Gerson and I. And VARIANT3D is a team of people with expertise in net engineering, quantum physics, game design, humane computing, VFX, architecture, computational geometry, aerospace and and many many others. And we are in the business of making software that knits.

0:32 And so let’s talk a little bit about knitting. Knitting is a technique that has been around for millennia. And the image I’m showing right here is an artifact of a knitted pair of socks from Egypt in 250 AD. And generally most of us today understand knitting in terms of this tradition, right? Two hands, two needles, and a yarn. And the long-standing technique and craft of hand knitting to me is is incredibly beautiful.

0:53 And it also speak to human ingenuity because just using our our mental faculty, we were somehow able to turn this conceptually one-dimensional material yarn into two-dimensional instruction. We call it a knit map or a knitting chart. And finally into a three-dimensional artifact, which is the fabric itself. And so there there’s something deeply arcane and alchemical about the the art of knitting. And it has been argued that knitting and weaving is actually due to its procedural nature was the prelude to modern computation as we know it today.

1:28 And so it’s it’s very appropriate too that now we also think of knitting in terms of its motorized computerized version. Machine knitting is the supercharged accelerated and industrial complement of hand knitting. And this technique has actually produced so much of our soft goods today. Like think about your sweaters or the your your your footwear uppers right in medical apparatus as well like we have custom fitted bandages and braces and even aerospace actually there’s a research in making knit engineered biosuit by MIT published in 2022 2023 and those are only a few example right but what exactly is a knitting machine so I’m going to show a schematic of one so this is a V-bed knitting machine as you can see it has feeder inputs on the left and the right and you have this carriage at the center that zips left and right, picks up and carries with it the feeders of yarns.

2:18 And while it while it’s doing that, there’s an actuator in the middle of this carriage that would activate the needles in the bed depending on the given machine instruction. And your fabric then would come out like right at the center of that machine gripped by the takedown. And if you zoom into the center, there’s hundreds of needles there. And this is how the yarns themselves are actually knitted.

2:41 It’s it’s a really beautiful machine. As you can see, it’s very waveike. It’s almost musical in nature. Almost rhythmic, right? Orchestral. You can think of the gesture of describing this this yarn carriers going to the left and the right as a conductor instead of saying, “Okay, go to this way and these are the notes they’re going to play.” But what’s interesting is at every pass, those performance actually occurs to become the fabric that comes out.

3:00 And so in thinking about what this manufacturing process affords, even though it has a number of topological constraints, you you can do so much with this technique. First of all, you can arrange the structure as you can see on the left. This is an image of Nike Air Zoom shoe. You can generate custom knit topology that amounts to different knit structures. So this is the equivalent to having different lises.

3:21 And second, you can generate a geometry. So this is a lamp cover that we developed at VARIANT3D. You can fabricate different shapes with zero waste. Your fabric comes out with full curvature, right? Coming from the machine and it also has the structure. And lastly, you can also compose with different materials. 16 different options from a from a standard Vbad knitting machine. The the right hand side is actually a project by MIT self assembly lab.

3:46 It’s called no foam knit. And the monofilament yarns that you see at the center there essentially gives it a very specific physical property. And so just looking at this image alone the potentiality of a knitting of of of knitting machines as an additive manufacturing tool is is absolutely immaculate. However, in the current process of designing knits, there’s there’s one very core challenge among many and what you’re seeing here is a single design cycle between designers on the left and a knit engineer on the right.

4:19 So let’s take the example of of a shoe design today. A designer would typically communicate in terms of diagrammatic patterns or other CAD models, right, with various callouts of desired properties. And on the other side, there’s this this deeply esoteric language that is well understood by by NIT engineers who have worked on it for many many years. And it’s essentially a gateway to transmute that design into a series of instructions on the machine.

4:45 And and on top of this alchemy, NIT engineers also exert some form of of geometric clairvoyance to to to intuit it how this two-dimensional net map would then translate into a threedimensional object. And all the while I want to I want to pinpoint this little thing at the center of the diagram, the prototype, the physical prototype. Designers and engineers have to speak in terms of that physical thing.

5:07 And we might ask ourselves, why is that? Why does the design iteration have to rely on that physical product? So the reason is twofold. Number one, there’s a discrepancy of language. So as a designer, I’d like to be able to specify that I desire X quality in my fabric. And on the other side, there’s a discrepancy of technical understanding. So as an engineer, I’d like to be able to effectively communicate that X is not achievable unless I make the compromise of Y.

5:34 So what we’re seeing here is a negotiation process. And one takeaway that I had from having been deeply embedded in the product development process at VARIANT3D is that there’s a lot of interpretation and negotiation that happened and within this design space itself this cycle is is very expensive right it takes 21 days for someone to be able to to say I would like to change my design I would like to alter my design because it doesn’t come out in the way that I intended or desire it to be and the design cycle here as you can see is also constrained by the availability of a net engineer they become a mechanical dependency to iterate on that output.

6:09 And on top of that, there’s also this combinatoric complexity of all the other variables, right, that influences the decided production strategy of that fabric. So, we’ve looked at a few. We talked about we talked about design, we talked about shape, materials, machine, but but what about the rest? You have cost of goods, you have different stakeholders, context of production, and we haven’t even talked about product varants such as grading and sizes.

6:34 So this space becomes very expensive very fast in terms of its logistical and temporal cost. Now the the question that we can ask ask ourselves after seeing this diagram is if this iteration is made somehow 10 orders of magnitude cheaper, what then changes, right? What kind of new design spaces can we actually afford? What kind of new products can we create? And what kind of new process, what kind of new system can actually emerge from it?

7:02 And answering this question is at the heart of our software loop. Loop is a software that knits and it’s for anyone to knit any shape with any complexity with with any machine anywhere in the world. It’s it’s centered around this idea that it’s possible to actually package the extremely complex system of knit engineering into intuitive design affordances almost like a painterly gesture. And this brush stroke can then be materialized with yarns.

7:27 And I’ve recorded a live demo for today’s presentation and I would like to show you how fast a single iteration can be with loop. The philosophy of loop is simple. If you have a browser and you can draw, you can knit. Start your object with a surface. We support any surface format and powered by unit gradient field and differential geometry. We allow our users to assign directional parameterization on the input surface.

7:55 And in the spirit of negotiation, designers can play in terms of target shape. Target knit structure and target materials. And afterwards, our yarn patter kernel, our yarn pattern in the kernel will return the machine ready output of that negotiation and the topologically accurate representation of that fabric. So what we’ve done here with loop is that we’ve allowed designer to design in a space in a in a context of affordance that is humane.

8:26 Right? Because this is how we think about products. We think about them in 3D. We think about them as physical things. But say what if you don’t want to draw freehand. Here I’m showing you that you can import a Rhino file and we’ll be supporting more CAD imports in the future. But loop and VARIANT3D as well actually we we value interoperability because many designers already have designs defined elsewhere.

8:47 And with this exact curves that we’re importing from Rhino look at how beautiful the net map is. This is very rhythmic. It’s very regular. So the machine files there’s an order of repetition that happens with it. And so what we’ve just seen in that presentation is in roughly 2 minutes we have transmuted this highle abstraction from shape from elsewhere into machine ready instruction and it’s opening up new design sensibilities.

9:22 So from shape into machine we import we parameterize we testate we do a yarn pathing and then we compile it into an agnostic knit code at the very bottom. And what are some other things that you can do with loop? So in a shoe this is a shoe designed by our team and loop also enables designers to define colorways and get virtual lookie samples. So you don’t have to knit them before you understand how your fabric is going to look like.

9:47 And the difference with other rendering methods is that these Lucy variants are actually compilable and they’re also knitable, right? And in terms of variance, another important feature that I would like to show today is grading. And it’s especially important in shoes. So you shouldn’t design 14 times in order to make 14 different sizes of a product. Not considering, you know, sometime you also have the narrow variants, you have the wide variants, many anatomical variants.

10:11 Rather you design once and interpolate the rest. And how is it that we handle precision with all of these different sizes? Loop have an automated calibration pipeline that will derive measurements from the fabric scans. We use state-of-the-art techniques such as Korean point drift. And we also support scanning of your fabric with any 3D scanners available. So photoggramometry, structured light scanning, and we’re even contemplating lumaf field and such.

10:37 And our kernel will make adjustment to your fabric. So the second one will always come out perfect based on the diffusion factor that is actually derived from this 3D scans. And so what we’ve seen is that we are able to show multiple colorways, multiple sizes, calibrated fit direct to machine. Now what about complex parametric geometry? This is a minimal surface with CDFM imprints on it. And I’m just going to cycle through some designs, some image imports that we can that we can actually use as drivers for knitting structures.

11:09 And one of them is called knit section is a photograph of a knitted fabric that we then imported as drivers. And Luke supports Floyd’s Steinberg dithering in 3D for up to five yarns. And imagine having to draw this by hand. That’s how it’s being done today. There’s just no way. And another interesting product I’d like to share is this one piece Laboo jumpsuit. Because who would have thought you can make one right and here I would like to show how we can use computational language also as a driver for the generation of knitting patterns now our system for isomorphism the transmuter powered by omega by gradient control laboratory it is a complement to the gestural interactive design so you don’t you don’t necessarily need to think of it in terms of a brush stroke anymore now we can start to think about emergent phenomena using our pyonic DS SL and we also enable our users to access new attention context that is actually powered by computers and the implication of this is that as designers you can start to think about one of one products right because the iteration cycle is so so small you can think about non-f fungeible products and lastly one thing I want to note here is by by aggregating the the knowledge of knitting inside our kernel and and using computational designs as as as drivers for patterns.

12:37 We also make it possible for for net design to be accessed by non-human agents, which is definitely not possible if you have this knowledge only in the silo of human brains. And so we would really like you to challenge us with your most complex geometry and bring in your parametric surfaces and then convert it into a knitted artifact that goes directly into the machine or code your patterns in loop and then feel it as a real fabric.

13:05 And in loop, we also enable the generation and iteration of products that would previously take years years. I’m talking years to develop. Performance needed bra, for example, is a new generation of apparel that is driven by by form biomechanics and and also the anatomy of movements, right? And it currently takes about 6 months to produce using traditional knit engineering techniques. But lube changes this. Now product designers can be a lot more nimble with their design and we can also start to infuse and and consider mechanical properties as drivers in the design process.

13:39 And some other adjacent applications in this field would include things like braces, bandages and compression garments that require very specific mechanical properties in in different area of the body. Now, one final product I would like to show that we’ve tried out with Loop is an automotive car seat. So, this is a product that previously take years to develop and perfect with months and months of iteration cycles.

14:01 And just looking at this, I think we can see that the loop is a new category of generative design software that essentially says that every surface you can imagine is now a computational canvas and you can make it manifest. You can make it into yarns that you can feel. And this computational canvas is also iterative. It’s collaborative and continuously emerging, right? And so I’d like us to imagine just using this workflow, what kinds of new soft goods can we actually make in the future?

14:36 So what are the surfaces that you’d like to get that you previously haven’t thought of? And so Loop is a software that enables design and computation to be made manifest. In the loop model, designers are able to produce a thousand times more design. And notice how NIT engineers here are entering the equation from an oblique angle, they are now given an an elevated role, right? Which is rather than being a mechanical dependency in a single iterative design cycle, they can now start to be the vanguard to opening complex design and techniques within our ecosystem.

15:13 And by doing so, we’re removing the barrier to deploying textiles at scale. And there are also a new emerging system that comes from this initial entry point. Loop now becomes a digital tech pack that knits all of this complexity. And you speak in terms of the same language. And in this ecosystem, because we’re now relying on a 21-day cycle, ideas will grow more ideas will grow and multiply exponentially.

15:40 And there is also a new kind of intuition that we’re building. And it’s an intuition towards a system and it’s a system that enables the production of of knitted goods. What we have oh sorry that got messed up but what we have also successfully done here is we have created a software that exudes the complex and combinatoric relationship between all this field of expertise. And at at our company we actually pride ourselves in being culturally and technically diverse.

16:08 And some might ask how how do we how do we get here? And the the answer to that is that Bryant 3D as an organization is is more than just software. And in almost a literal manner, it is also a destination. When VARIANT3D was first started by Garrett, he decided that the office is going to be situated in the middle of California’s renowned resort, Calamigos. And and Garrett on top of his visionary leadership in the in the industry I think has in a very unique and special way successfully infused this compound as a core part in our culture as a company and our quarterly team trip is always a good time and this is our HQ it is a repurposed horse barn at the center of the resort and this is our protoing lab as well and it is exactly in this compound where VARIANT3D invites and works with the world’s leading thinkers, brands, designers to make the world’s most advanced knitted product.

17:05 And this is where we build relationship and empathy with our clients and users. And some people here are going to be in our HQ next week. So, looking forward to seeing you there. And it is also here where future NIT production system is actually incubated and made and made real, right? So, our protoing lab has been 8 years in practice. We have over a 100 person year of knit engineering expertise.

17:33 More than 10 knitting machines. We have 3D printers. We have 3D scanners. We also offer technology accelerated guided development services for people who desire to make really advanced knitted products and a robust customer support. And we’re actively collaborating with international factories such as Apache and AIT in order to usher in this new system of of manufacturing and continuously collaborating with academic institutions such as MIT self assembly lab.

17:59 We’re working actively with Skyler and Boston University Northeastern and Carnegie Melon. And so we’re actively collaborating because there are many many ways that you can approach Ned and the only way I think we can get to to to to a robust ecosystem together is to share our ideas. And so to conclude my presentation I think we’ve talked about three things today. We have material, we have software and we have systems.

18:26 And the core question here is what are their humane features? So as for system change design iteration process within a system and make it intuitive and all sorts of beautiful things will emerge from it. You’re going to start to see new language, new products and also new workflows. And as for software, software is never done. And for the very reason that it is never done, it becomes this this sponge this this sort of coral that absorbs the context where where it is situated, right?

18:54 And and loop with VARIANT3D will always continue to exude the relationship that we’ve built with our clients, the practice that we’ve built as a company and most importantly will continue to contain multitude of possibilities. It’s never close-ended. And lastly, as for material, there’s this beautiful material and an powerful one in this world called yarn. And it’s it’s so powerful that it makes almost any surface that you see a shape shifter and kind of a pool of possibilities.

19:20 And so imagine the kinds of soft goods that will rest on our skin, wrap around our feet, texture our lift environments, and ultimately augment our thoughts and identities as as humans. So the material potentiality of textile is a deeply magical one for me and also for everyone at VARIANT3D. And we invite all of you here to join us in bringing them to life. And if you’d like to get in contact, we also just launched our new website, varant3d.io.

19:48 We’re releasing our private beta on the 12th of January 2026 to select enterprise clients. Some of the biggest brands you can imagine that I I cannot name. And we’re backed up with high budget customers in the next few months. But anyhow, come and chat with us. Go to the top right. Get early access and reach out. Send us your craziest cat cards or your wildest idea. Let’s see what we can make happen together.

20:09 Say, thank you for your time. And here’s my digital business cards in the lab. And come say hi. Don’t be a stranger. Happy to give anyone a live demo after this presentation. Thank you. Materials as well. Yes. Seems like a zitic. Yes, exactly.